📚 What this platform does
This system is operated by AABE Coalition in Taiwan, integrating three civil society watchdog platforms — one per UN treaty body — under a shared technical backbone (evidence_pipeline.py):
- Crawls government websites publishing reports, NAPs, regulatory reviews, etc.
- Extracts clean text via BeautifulSoup main-content selectors per site
- Indexes 100K+ passages with SQLite FTS5 full-text search
- Auto-links policy issues (PIs) to evidence using keyword matching
- Generates frequency time-series for tracked terminology over years
Each platform retains its own issue framework, advocacy stance, and host organization. The shared system enables cross-covenant comparison — finding intersectional gaps invisible to any single covenant.
♀ CEDAW (Women's Rights)
Host: Taiwan Mommies Family & Children Coalition
Acceded 2007 · 4 review cycles · 5th in preparation (2026)
→ CEDAW public site| Indicator | Count |
|---|---|
| Documents | 394 |
| Passages indexed | 29,638 |
| Policy Issues (PI) | 18 |
| PI ↔ evidence linkages | 424 |
| Concluding Observations | 212 |
| Stakeholders mapped | 63 |
☻ CRC (Child Rights)
Host: AABE Coalition for Education Action
Acceded 2014 · 2 review cycles · 3rd in progress
→ CRC public site| Indicator | Count |
|---|---|
| Documents | 485 |
| Passages indexed | 69,231 |
| Policy Issues (PI) | 14 |
| PI ↔ evidence linkages | 714 |
| Concluding Observations | 170 |
| Stakeholders mapped | 56 |
⚙ CRPD (Disability Rights)
Host: Disability NGO coalition (in formation)
Acceded 2014 · 2 review cycles · 3rd in preparation
→ CRPD public site| Indicator | Count |
|---|---|
| Documents | 1,168 |
| Passages indexed | 3,298 |
| Policy Issues (PI) | 16 |
| PI ↔ evidence linkages | 171 |
| Stakeholders mapped | — |
🔬 Cross-covenant findings (highlights)
Selected findings from the shared analytical platform — invisible to any single covenant NGO:
| Topic | CEDAW | CRC | CRPD | Insight |
|---|---|---|---|---|
| Reasonable accommodation (合理調整) | 100 | 173 | 7 | CRPD's core concept appears least in CRPD government documents — structural under-representation |
| Disabled women (CEDAW × CRPD) | 190 | 3 | 0 | Despite CEDAW GR-18 on disabled women, no CRPD government document discusses this intersection |
| Domestic violence (家庭暴力) | 459 | 515 | 1 | Disabled women's intersectional vulnerability to violence absent from CRPD policy discourse |
| SOGIESC framing (LGBTI/性別認同/etc.) | 1,491 | 418 | 297 | Concentrated in women's policy (CEDAW); 100× higher rate of expansion than CRC/CRPD |
| Disabled indigenous (3-covenant intersection) | 0 | 1 | 0 | Triple-axis intersection completely absent across all government discourse |
🌐 Why integrate three watchdogs?
Single-covenant NGOs structurally cannot detect these gaps. A women's rights NGO does not regularly read CRPD government reports; a disability rights NGO does not survey CEDAW frequency. Only when datasets are unified can we identify topics that every covenant treats as "the other's responsibility" — what UN human rights mechanisms call intersectional gaps.
This platform provides civil society and treaty body reviewers with empirical evidence of:
- Where government discourse converges or diverges across covenants
- Which intersectional populations (disabled women, LGBTI children, indigenous women, etc.) are systematically under-discussed
- How tracked terminology (LGBTI, "reasonable accommodation", etc.) evolves over time per covenant
- Which civil society advocacy events have triggered policy changes (or have not)
🛠 Open-source infrastructure
evidence_pipeline.py — a 900-line Python module with 8 classes (Fetcher, MarkdownExtractor, SiteCrawler, DocumentImporter, PassageIngester, EvidenceLinker, FrequencyAnalyzer, EvidenceInjector). Same MD5 across 3 platforms. Per-platform configuration via PipelineConfig dataclass.
Future planned open-sourcing: github.com/naer-tw/evidence-pipeline with full English documentation, inviting contributors from other covenants (ICCPR / ICESCR / CAT / CERD).